KR20080101014A - Method and apparatus for predicting radio channel in wireless communication system - Google Patents

Abstract

A method for predicting a wireless channel in a wireless communication system, and an apparatus thereof are provided to accurately calculate the beamforming or precoding weight of a transmitter through the prediction of a downlink channel using the Kalman filter in a TDD system. A method for predicting a wireless channel in a wireless communication system comprises the following steps of estimating the current uplink channel; predicting the current uplink channel from an estimated predetermined previous uplink; and predicting the next downlink channel from the uplink channel prediction. The process of predicting the current uplink channel from the estimated previous uplink is implemented by the following steps. The predetermined previous uplink channel predictive value is filtered to predict the current uplink channel. A parameter for correcting the predicted uplink channel is calculated. The uplink channel value is corrected by using the parameter.

Description

Method and apparatus for predicting wireless channel in wireless communication system {METHOD AND APPARATUS FOR PREDICTING RADIO CHANNEL IN WIRELESS COMMUNICATION SYSTEM}

1 is a frame structure in a broadband wireless communication system according to the prior art,

2 is a configuration diagram of an uplink system for sounding symbols in a broadband wireless communication system according to the present invention;

3 is a block diagram of a downlink precoding system in a broadband wireless communication system according to the present invention;

4 is a frame structure for channel prediction in a broadband wireless communication system according to the present invention;

5 is a detailed channel prediction apparatus in a broadband wireless communication system according to an embodiment of the present invention;

6 is an operation flowchart for channel prediction in a broadband wireless communication system according to an embodiment of the present invention;

7 is a graph of bit error rate (BER) performance when using a linear predictor and a Kalman filter according to an embodiment of the present invention;

8 is a graph of mean square error (MSE) of sounding symbols predicted when using a linear predictor and a Kalman filter according to an embodiment of the present invention;

9 is a graph showing bit error rate (BER) performance when using a linear predictor and a Kalman filter according to another embodiment of the present invention;

10 is a graph showing the mean square error (MSE) of sounding symbols predicted when using a linear predictor and a Kalman filter according to another embodiment of the present invention.

The present invention relates to a method and apparatus for predicting a wireless channel in a wireless communication system. In particular, the channel information of the uplink is filtered using a Kalman filter and the downlink channel is filtered using the filtered uplink channel information. A method and apparatus for predicting are provided.

In general, in a multi-antenna system, when a transmitter of a base station knows channel state information (hereinafter referred to as "CSI"), a higher channel capacity may be achieved using a technique such as beamforming. . In addition, the signal processing process for recovering the signal received from the multiple antennas in advance in the transmitter, it is possible to simply implement a low complexity in the terminal limited to the signal size and power consumption. As a method for obtaining channel information in the transmitter, a method of feeding back channel information in the receiver is common.

When the transmitter intends to obtain downlink channel information, a method of feeding back the downlink channel information obtained from the terminal to the transmitter is common. This method can be used not only in the TDD system but also in a frequency division duplex (FDD) system in which uplink and downlink are not related to each other. However, when considering a multi-antenna system, as the number of antennas increases, necessary feedback information also increases exponentially. On the other hand, in a time division duplex (TDD) system, the channel does not need to be fed back.

1 shows a frame structure in a broadband wireless communication system according to the prior art. One frame is divided into downlink and uplink, and is divided into transmission times. The uplink frame includes an uplink control symbol, a data symbol, and a sounding symbol. The sounding symbol may be selectively used to use a TDD characteristic as a symbol for a closed-loop Multi Input Multi Output (MIMO). The downlink frame includes a preamble, a frame control header (FCH), a data symbol, and the like.

Specifically, since the TDD system has a mutual property between uplink and downlink channels, it is possible to obtain downlink channel information from uplink using this. However, a time delay may occur due to the constraint that the uplink and the downlink are separated by time in the frame structure. In other words, the channel continues to change with time, but during downlink, the transmitter cannot obtain new channel information, but only delayed channel information obtained from the uplink of the previous frame. In order to reduce the error between the real channel caused by the time difference between uplink and downlink and the channel obtained from the base station, linear prediction can be performed using the channel information obtained from the uplink of the previous frame and the previous frame. have.

As described above, in the case of the linear prediction technique, in order to predict a downlink channel, a transmitter first estimates an uplink channel. In this case, since the estimated uplink channel includes noise, the estimated uplink channel has an error from the actual channel. When linear prediction is performed using the information including the error, the predicted downlink channel error becomes larger. This results in serious performance degradation when calculating beam beam weights or precoding weights in actual transmitters because they do not match the actual values.

Accordingly, there is a need for a method and apparatus for performing channel prediction by reducing an error between a real channel generated by a time difference between uplink and downlink and a channel obtained by a base station in a TDD-based wireless communication system.

Accordingly, an object of the present invention is to provide a method and apparatus for predicting a radio channel in a wireless communication system.

Another object of the present invention is to provide a method and apparatus for filtering uplink channel information and predicting a next downlink channel by using a Kalman filter in a TDD-based wireless communication system.

Another object of the present invention is to provide a method and apparatus for performing channel prediction by reducing an error between a real channel generated by a time difference between uplink and downlink and a channel obtained from a base station in a TDD-based wireless communication system.

According to a first aspect of the present invention for achieving the above objects, a method for estimating a radio channel in a wireless communication system, comprising: estimating a current uplink channel and a current uplink channel from the estimated previous uplink And predicting a next downlink channel from the current uplink channel prediction.

According to a second aspect of the present invention for achieving the above objects, in a wireless channel prediction apparatus, a channel estimator for estimating a current uplink channel, and a current uplink from the estimated previous uplink And predicting a channel to predict a next downlink channel from the current uplink channel prediction.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description of the present invention, detailed descriptions of related well-known functions or configurations will be omitted if it is determined that the detailed description of the present invention may unnecessarily obscure the subject matter of the present invention. Terms to be described later are terms defined in consideration of functions in the present invention, and may be changed according to intentions or customs of users or operators. Therefore, the definition should be made based on the contents throughout the specification.

Hereinafter, a method and apparatus for filtering uplink channel information and predicting a next downlink channel using a Kalman Filter in a time division duplex (TDD) based wireless communication system will be described. . The Kalman filter is used to find signals from noise so that one system can properly predict changes over time.

2 to 6 illustrate a broadband wireless communication system as an example, it is a matter of course that the present invention can also be applied to other types of TDD-based broadband wireless communication systems.

2 is a block diagram of an uplink system for sounding symbols in a broadband wireless communication system according to the present invention. The uplink system refers to a system in which a base station (receiver) receives the control signal or the data signal through a reception antenna when a terminal (transmitter) transmits a control signal or a data signal to a wireless channel through a transmission antenna.

Referring to FIG. 2, the apparatus for transmitting a terminal includes a sounding symbol generator 200, an IFFT unit 202, and a parallel-serial converter 204.

The sounding symbol generator 200 generates a sounding symbol and outputs the sounding symbol to the IFFT unit 202 so that the base station can perform uplink channel estimation. The IFFT unit 202 converts a sounding symbol from the sounding symbol generator 200 into an inverse fast Fourier transform (IFFT) to a signal in a time domain (referred to as an OFDM symbol). In this case, the IFFT unit 202 may perform a cyclic prefix (CP) insertion. The parallel-serial converter 204 converts the parallel data output from the IFFT unit 202 into serial data and transmits the serial data.

The receiving apparatus of the base station includes a plurality of serial-parallel converters 201 to 201_n, a plurality of FFT units 203 to 203_n, a channel estimator 205, and a channel predictor 207.

The serial-parallel converter 201 to 201_n converts the received sounding symbols from the terminal in parallel and outputs them to the FFT units 203 to 203_n. The FFT units 203 to 203_n perform fast Fourier transform on the data from the serial-parallel conversion units 201 to 201_n, and output data of the frequency domain to the channel estimator 205. In addition, the FFT units 201 to 201_n may perform CP removal.

The channel estimator 205 extracts a sounding symbol (corresponding to a pilot symbol of the downlink) from data from the FFT units 203 through 203_n and performs channel estimation to obtain a channel estimation result. Will output

The channel predictor 207 performs Kalman filtering to use the uplink channel prediction value corresponding to the previous sounding symbol (referred to as n-1 sounding symbol) to the next uplink channel (uplink corresponding to n frame). After predicting the link channel), the downlink channel corresponding to the n + 1 frame is predicted using the channel prediction value of the nth uplink. A detailed description of the channel predictor 207 will be described with reference to FIG. 5.

3 is a block diagram of a downlink precoding system in a broadband wireless communication system according to the present invention. The downlink system refers to a system in which a base station (transmitter) transmits a control signal or a data signal to a wireless channel through a transmission antenna, and the terminal (receiver) receives the control signal or the data signal through a reception antenna.

Referring to FIG. 3, the transmitter of the base station includes a plurality of modulators 300 to 300_n, a precoding unit 302, a plurality of IFFT units 304 to 304_n, and a plurality of parallel-to-serial conversion units 306 to 306_n. It is configured to include.

The modulators 300 to 300_n modulate input bits in a predetermined modulation scheme to generate modulation symbols and output the modulation symbols to the precoding unit 302. Here, the modulation scheme includes binary phase shift keying (BPSK) for mapping one bit to one signal point (modulation symbol), quadrature phase shift keying (QPSK) for mapping two bits to one modulation symbol, and three There are 10-ary phase shift keying (8PSK) for mapping bits to one modulation symbol, and 16QAM for mapping four bits to one modulation symbol.

The precoding unit 302 calculates a beam pome weight or a precoding weight with reference to a downlink channel prediction value from the channel prediction unit 207, and multiplies the modulation symbols from the modulators 300 to 300_n. Output to the IFFT unit 304 to 304_n. That is, the precoding unit 302 may minimize signal interference by pre-processing modulation symbols from the modulators 300 to 300_n, and radiating inter-symbol interference or directional beams.

The IFFT units 304 to 304_n and the parallel-serial converters 306 to 306_n perform the same functions as the IFFT units 202 to 202_n and the parallel-serial converters 204 to 204_n of FIG. 2.

The receiver of the terminal includes a serial-parallel converter 301, an FFT unit 303, and a demodulator 305.

The series-parallel conversion unit 301 and the FFT unit 303 perform the same functions as the series-parallel conversion unit 201 to 201_n and the FFT units 203 to 203_n of FIG. 2.

The demodulator 305 receives data from the FFT units 303 to 303_n, receives a soft output value, performs soft decision, outputs coded bits, and decodes the information bits according to the corresponding coding scheme. Output For example, if the transmitting end is encoded using a convolutional code, the receiving end outputs an information bit through a Viterbi algorithm.

4 illustrates a frame structure for channel prediction in a broadband wireless communication system according to the present invention.

Referring to FIG. 4, in the frame structure of the TDD system, uplink and downlink are divided into transmission times, and a sounding symbol is added at the end of the uplink frame. Since the detailed structure of the frame structure of the TDD-based broadband wireless communication system is not important in the present invention, a detailed description thereof will be omitted.

)한다. 여기서, 상기 n은 프레임 인덱스이고, 상기

은 필터링되어 예측된 채널을 의미한다. 마지막으로 세 번째 단계에서(404), 상기 TDD 시스템은 추정한 사운딩 심벌과 예측한 사운딩 심벌을 이용하여 선형보간(Linear Interpolation) 기법을 통해 n+1 번째 하향링크 채널을 예측하게 된다.In the above-described TDD system frame structure, a process of predicting a downlink channel using a Kalman filter is described. First, in a first step 400, the TDD system estimates an uplink channel using sounding symbols located at the end of each uplink frame. In the next second step (402), the TDD system predicts the sounding symbols of the frame by filtering the sounding symbols estimated by the Kalman filter on the uplink channels estimated in the previous frame (

)do. Where n is a frame index and

Means filtered and predicted channel. Finally, in the third step (404), the TDD system predicts the n + 1th downlink channel through linear interpolation using the estimated sounding symbol and the predicted sounding symbol.

라 하면, n 번째 프레임에 대한 채널응답을 하기 <수학식 1>과 같이 AR 모델로 표현할 수 있다.In order to describe Kalman filtering and channel prediction in more detail in FIG. 5, the channel response will be described by expressing it as an AutoregRessive (hereinafter referred to as "AR") model. The k-th channel response of the sounding channel of the p-th transmit antenna and the q-th receive antenna

In this case, the channel response for the n-th frame may be expressed by an AR model as shown in Equation 1 below.

는 차수(order)가 u인 AR 모델의 계수이고, 상기

은 평균이 0이고 분산이 상기

인 백색 가우시안 잡음이다. 채널의 상관관계(Correlation)로부터 율-워커 방정식(Yule-Walker equation)을 풀면 상기

와 상기

를 할 수 있다.Where

Is a coefficient of an AR model of order u, and

Has an average of 0 and the variance

It is a white Gaussian noise. Solving the Yule-Walker equation from the correlation of the channel

And above

You can do

을

라고 정의하면, 칼만 필터를 위한 상태 방정식(state equation)과 관측 방정식(Observation Equation)은 하기 <수학식 2>, 하기 <수학식 3>으로 각각 표현할 수 있다. 여기서, T는 전치행렬이고, u는 AR 모델의 차수이고, n은 프레임 인덱스이다.

of

In this case, the state equation and the observation equation for the Kalman filter can be expressed by Equation 2 and Equation 3, respectively. Where T is the transpose matrix, u is the order of the AR model, and n is the frame index.

는 차수가 u인 AR 모델에서 따른 채널응답 벡터이고, 상기

는 차수가 u인 AR 모델 계수이고, 상기

은 백색 가우시안 잡음이다. 즉, 상기

는

이고, 상기

는

으로 표현된다.Where

Is the channel response vector according to the AR model of order u,

Is an AR model coefficient of order u, and

Is a white Gaussian noise. That is

Is

And

Is

It is expressed as

는 상향링크 사운딩 심벌을 이용한 계산된 채널추정 값으로 칼만 필터의 측정 데이터로 사용되고, 상기

는 측정 행렬(measurement matrix)이고, 상기

는

으로 잡음 벡터이다.Where

Is a channel estimation value calculated using an uplink sounding symbol and is used as Kalman filter measurement data.

Is a measurement matrix, and

Is

Is a noise vector.

5 is a diagram illustrating a detailed channel prediction apparatus in a broadband wireless communication system according to an embodiment of the present invention.

The channel predictor 207 includes a time updater 500, a measurement updater 502, a channel corrector 504, and a channel predictor 506.

The time updater 500 calculates the predicted uplink channel value of the current frame by using the filtered information of the previous frame (predicted channel value of the previous frame) and outputs the estimated uplink channel value of the current frame to the measurement updater 502. For example, when the order of the AR model is 2, the channel of the n th current uplink frame is predicted using the predicted channel values for the uplink frames n-1 and n-2. If the AR order is m, the predicted channel values for the m previous uplink frames are used to predict the current uplink frame.

In addition, the time updater 500 calculates an error covariance used to minimize a mean square error and outputs the error covariance to the measurement updater 502.

)과 에러 공분산(

)은 하기 <수학식 4>로 표현된다.Channel prediction value of the n th current uplink frame (

) And error covariance (

) Is expressed by Equation 4 below.

는 n-1 번째의 상향링크 프레임 채널예측 값을 이용하여 산출된 n 번째 상향링크 프레임 채널예측 값이고, 상기

는 AR 모델 계수 값이고, 상기

는 n-1 번째의 에러 공분산을 이용하여 산출된 n 번째 에러 공분산이고, 상기

는 잡음이다. 이때, 상기

는 제로벡터로,

은 단위행렬로 초기화한다. 즉,

과

의 초기화 값은 각각

,

이다. 그리고,

이다.Where

Is the n-th uplink frame channel prediction value calculated using the n-th uplink frame channel prediction value.

Is an AR model coefficient value, and

Is the n th error covariance calculated using the n-1 th error covariance,

Is noise. At this time, the

Is a zero vector,

Is initialized to the unit matrix. In other words,

and

The initial value of each is

,

to be. And,

to be.

,

를 제공받아, 채널보정을 위한 파라미터를 계산하여 상기 채널보정기(504)로 출력한다. 여기서, 상기 채널보정을 위한 파라미터인 칼만 이득(

), 이노베이션 벡터값(

)은 하기 <수학식 5>로 산출된다.The measurement updater 502 is provided from the time updater 500.

,

Received to calculate the parameters for channel correction and outputs to the channel corrector (504). Here, the Kalman gain (parameter for the channel correction)

), The innovation vector value (

) Is calculated by Equation 5 below.

은 n 번째 프레임의 칼만 이득이고, 상기

는 n-1 번째의 에러 공분산이고, 상기

는 측정 행렬이고, 상기

는 잡음 벡터이고, 상기

은 n 번째 프레임의 이노베이션 벡터값이고, 상기

는 상향링크 사운딩 심벌을 이용한 계산된 채널추정 값이고, 상기

는 n-1 번째의 상향링크 프레임 채널예측 값이다. 그리고,

이다.Where

Is the Kalman gain of the nth frame,

Is the n-1 th error covariance,

Is the measurement matrix, and

Is a noise vector, and

Is the innovation vector value of the nth frame,

Is a calculated channel estimation value using an uplink sounding symbol,

Is the n-1 th uplink frame channel prediction value. And,

to be.

The channel corrector 504 corrects the uplink channel prediction value and the error covariance of the current frame by using the parameter for channel correction from the measurement updater 502. Here, the uplink channel prediction value and the error covariance of the corrected current frame are expressed by Equation 6 below.

은 보정된 n 번째의 상향링크 프레임 채널예측 값, 상기

는 n-1 번째의 상향링크 프레임 채널예측 값을 이용하여 산출된 n 번째 상향링크 프레임 채널예측이고, 상기

은 n 번째 프레임의 칼만 이득이고, 상기

은 n 번째 프레임의 이노베이션 벡터값이다. 상기

는 보정된 n 번째의 에러 공분산이고, 상기

는 단위벡터이고, 상기

는 측정 행렬이고, 상기

는 n-1 번째의 에러 공분산을 이용하여 산출된 n 번째 에러 공분산이다.Where

Is a corrected n-th uplink frame channel prediction value,

Is the n-th uplink frame channel prediction calculated using the n-th uplink frame channel prediction value.

Is the Kalman gain of the nth frame,

Is the innovation vector value of the nth frame. remind

Is the corrected n th error covariance,

Is a unit vector, and

Is the measurement matrix, and

Is the n th error covariance calculated using the n-1 th error covariance.

The channel predictor 506 receives the uplink channel prediction value of the current frame corrected by the channel corrector 504 to predict the next downlink channel.

은 n+1 번째 하향링크의 채널예측값이고, 상기

는 AR 모델 계수 값이고, 상기

은 보정된 n 번째의 상향링크 프레임 채널예측 값이다.

은 x 벡터의 첫 번째 성분을 나타낸다.Where

Is the channel prediction value of the n + 1 th downlink,

Is an AR model coefficient value, and

Is the corrected n-th uplink frame channel prediction value.

Denotes the first component of the x vector.

와 예측된 채널

사이를 선형보간 기법에 의해 최종적으로 하향링크의 채널정보를 예측한다.As described above, the channel predictor 207 predicts an uplink sounding channel of the next frame through one-step prediction and correction through Kalman filtering. Filtered channels

And predicted channels

Finally, downlink channel information is predicted by linear interpolation.

채널을 추정하는 동안

프레임이 지나가 버리는 상황이 발생하는 경우,

,

,

으로

을 추정하는 것이 아니라 하나 건너서

를 예측한다. 이 경우 (n+1) 프레임에서의 측정 데이터가 없기 때문에 곧바로 (n+2) 시간에서의 채널을 예측할 수 없다. 따라서, (n+1) 시간에서의 측정 데이터를 예측해야만

를 예측할 수 있다.

를 구하기 위해 먼저, (n+1) 프레임에서의 측정 데이터를 예측한다. 필터링한 채널

과

을 선형 예측하여 얻은 값을 (n+1) 프레임에서의 측정 데이터로 사용한 다음, 위에서 예측한 채널

과 함께 칼만 필터로

를 구한다.Meanwhile,

While estimating the channel

If a situation occurs where a frame passes by,

,

,

to

Rather than estimating one

Predict. In this case, since there is no measurement data in the (n + 1) frame, the channel at (n + 2) time cannot be immediately predicted. Therefore, the measurement data at (n + 1) time must be predicted

Can be predicted.

First, the measurement data in the (n + 1) frame is predicted. Filtered channels

and

Is used as the measurement data in the (n + 1) frame, and then the channel predicted above

With Kalman Filter

Obtain

6 is a flowchart illustrating an operation of channel prediction in a broadband wireless communication system according to an exemplary embodiment of the present invention.

은 제로벡터로, 에러 공분산 행렬

은 단위행렬로 초기화한다. First, the receiver of the base station receives the sounding symbol from the terminal in step 600 and initializes the state equation to predict the uplink channel. For example, the filtered channel prediction matrix

Is a zero vector, the error covariance matrix

Is initialized to the unit matrix.

Thereafter, the receiver estimates the channel using the sounding symbol of the terminal received through the reception antenna in step 602.

Thereafter, the receiver updates the time using the frame of the previous sounding symbol in step 604. Herein, the frame of the previous sounding symbol is referred to as n-1 frame for the sake of convenience. That is, in step 604, the uplink channel prediction and the error covariance of n frames are updated using the filtered information of the n-1 frames (the predicted channel value of the previous frame).

, 이노베이션 벡터(Innovation Vector)

등이 있다.In step 606, the receiver calculates a parameter for channel correction and performs a measurement update. The parameter for channel correction is Kalman filter gain.

, Innovation Vector

Etc.

In step 608, the receiver predicts an uplink channel for n frames by performing channel correction using a parameter for channel correction.

In step 610, the receiver predicts the downlink channel for the n + 1 frame using the uplink channel for the n frame using the prediction value.

Thereafter, the receiver determines whether channel estimation is necessary in step 612, and if channel estimation is necessary, repeats steps 602 to 610 to continue channel prediction for the next frame.

If the receiver does not need channel estimation in step 612, the receiver terminates the algorithm of the present invention.

7 to 10 show a simulation result graph according to an embodiment of the present invention.

First, describing the simulation environment, the following <Table 1> shows the simulation system parameters. The system variables related to OFDM symbol and frame structure for simulation are experimented based on broadband wireless communication system (eg IEE802.16e).

Carrier frequency fc = 2.3 GHz Sampling frequency fs = 10 MHz Frame length Tfr = 5ms OFDM symbol length (including CP) Tsym = 115.2 μs Receiver speed 3km / h Normalized Doppler Frequency for Sounding Symbol fdTfr = 0.0319 OFDM symbol per frame (DL / UL) 42 (NDL = 27 / NUL = 15) FFT size N = 1024 CP length L = 128 Modulation uncoded QPSK Antenna Number (BS, MS) (Nt = 2, Nr = 2) Channel model (fading / multipath) Jakes fading / ITU-R PED-B AR model order 1, 2, 3

FIG. 7 shows bit error rates (hereinafter, referred to as "BER") when the channel prediction is not used, when the linear prediction is performed, and when the channel prediction is performed using the Kalman filter. Signal-to-noise ratio (hereinafter referred to as "SNR") of the uplink is 9dB, the precoding method is a channel inversion (Zero-forcing) method was used. Here, the power limitation of the transmitter is not considered and the receiver does not do any work for channel compensation. That is, an uncoded QPSK signal is multiplied by a precoding weight obtained from a transmitter and received by a receiver through a radio channel to demodulate. As shown in FIG. 7, when the Kalman filter is used, an error due to channel delay of uplink and downlink can be reduced to improve performance. When the order of the Kalman filter is 1, since the modeling is much different from the actual channel when AR modeling, the performance is lower than the case of using a constant precoding weight without channel prediction. On the other hand, when the degree of the filter is 3, the performance is almost the same as when the degree is 2.

의 평균제곱오류(Mean Square Error: MSE)를 상향링크의 SNR에 대하여 나타내고 있다.8 shows a sounding channel predicted by a linear prediction method and a prediction method using a Kalman filter,

Mean Square Error (MSE) is shown for the uplink SNR.

,

,

으로

을 추정하는 것이 아니라 하나 건너서

를 예측했을 경우의 BER 성능을 보여준다. 이 경우 상기 도 8에서처럼 성능 개선을 가져올 수 있다.9 is

,

,

to

Rather than estimating one

It shows the BER performance when we predict. In this case, as shown in FIG. 8, performance may be improved.

의 평균 MSE를 상향링크의 SNR에 대하여 나타내고 있다.10 is a sounding symbol predicted across one

The average MSE of is shown for the uplink SNR.

As can be seen from the simulation results, the beamforming or precoding weight of the transmitter can be more accurately calculated through the downlink channel prediction using the Kalman filter in the TDD system, and the performance can be improved accordingly.

The present invention can be applied to not only a wideband wireless communication system but also other TDD systems to expect the same performance improvement.

Meanwhile, in the detailed description of the present invention, specific embodiments have been described, but various modifications are possible without departing from the scope of the present invention. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined not only by the scope of the following claims, but also by the equivalents of the claims.

As described above, the beamforming or precoding weight of the transmitter can be more accurately obtained through downlink channel prediction using a Kalman filter in the TDD system, and performance can be improved accordingly. In other words, by predicting a channel changing during downlink, the transmitter can reduce the error in calculating the beam pome weight or the precoding weight, thereby improving performance.

Claims (22)

In the wireless channel prediction method in a wireless communication system, Estimating the current uplink channel; Predicting the current uplink channel from the estimated previous uplink; Predicting a next downlink channel from the current uplink channel prediction. The method of claim 1, Predicting a current uplink channel from the estimated previous uplink Predicting a current uplink channel by filtering the predetermined previous uplink channel prediction value; Calculating a parameter for correcting the predicted current uplink channel; And correcting a current uplink channel value using the parameter. The method of claim 2, And filtering the predetermined previous uplink channel prediction value using a Kalman filter. The method of claim 2, The parameter for correcting the predicted current uplink channel is Kalman gain (

), The innovation vector value (

). The method of claim 4, wherein The Kalman gain and the innovation vector value is calculated by the following Equation (8).

Where

Is the Kalman gain of the nth frame,

Is the n-1 th error covariance,

Is the measurement matrix, and

Is a noise vector, and

Is the innovation vector value of the nth frame,

Is a calculated channel estimation value using an uplink sounding symbol,

Is n-1th uplink frame channel prediction value. The method of claim 4, wherein The corrected uplink channel value is calculated by Equation 9 below.

Where

Is a corrected n-th uplink frame channel prediction value,

Is the n-th uplink frame channel prediction calculated using the n-th uplink frame channel prediction value.

Is the Kalman gain of the nth frame,

Is the innovation vector of the nth frame. The method of claim 1, The current uplink channel estimation method using uplink sounding symbols. The method of claim 1, And a predetermined number of previous uplink channel prediction values used to predict the current uplink channel is determined by autoregressive (AR) model coefficients. The method of claim 1, The predicted next downlink channel is calculated by the following equation (10).

Where

Is the channel prediction value of the n + 1 th downlink,

Is an AR model coefficient value, and

Is the corrected nth uplink frame channel prediction value. The method of claim 1, and when the n + 1 th downlink channel prediction time passes while estimating the n th downlink channel, the n + 2 th downlink channel prediction is performed, and the measurement data of the predicted n + 1 downlink channel is used. The method of claim 1, The next downlink channel prediction from the current uplink channel prediction is performed in a TDD-based Multi Input Multi Output (MIMO) Orthogonal Frequency Division Multiplexing (OFDM) system. In the wireless channel prediction apparatus in a wireless communication system, A channel estimator for estimating a current uplink channel; And a channel predictor for predicting a current uplink channel from the estimated previous uplink and predicting a next downlink channel from the current uplink channel prediction. The method of claim 12, The channel predictor A time updater for predicting a current uplink channel by filtering the predetermined previous uplink channel prediction value; A measurement updater for calculating a parameter for correcting the predicted current uplink channel; And a channel corrector for correcting a current uplink channel value using the parameter. The method of claim 13, And the Kalman Filter is used to filter the predetermined previous UL channel prediction value. The method of claim 13, The parameter for correcting the predicted current uplink channel is Kalman gain (

), The innovation vector value (

Device). The method of claim 15, And the Kalman gain and the innovation vector value are calculated by Equation (8).

Where

Is the Kalman gain of the nth frame,

Is the n-1 th error covariance,

Is the measurement matrix, and

Is a noise vector, and

Is the innovation vector value of the nth frame,

Is a calculated channel estimation value using an uplink sounding symbol,

Is n-1th uplink frame channel prediction value. The method of claim 15, The corrected current uplink channel value is calculated by the following equation (9).

Where

Is a corrected n-th uplink frame channel prediction value,

Is the n-th uplink frame channel prediction calculated using the n-th uplink frame channel prediction value.

Is the Kalman gain of the nth frame,

Is the innovation vector of the nth frame. The method of claim 12, Wherein the current uplink channel estimation uses an uplink sounding symbol. The method of claim 12, And a predetermined number of previous uplink channel prediction values used to predict the current uplink channel is determined by an AutoregRessive (AR) model coefficient. The method of claim 12, Wherein the predicted next downlink channel is calculated by Equation (13).

Where

Is the channel prediction value of the n + 1 th downlink,

Is an AR model coefficient value, and

Is the corrected nth uplink frame channel prediction value. The method of claim 12, and when the n + 1 th downlink channel prediction time passes while estimating the n th downlink channel, the n + 2 th downlink channel prediction is performed, and the measurement data of the predicted n + 1 downlink channel is used. The method of claim 12, And a next downlink channel prediction from the current uplink channel prediction is performed in a TDD-based Multi Input Multi Output (MIMO) Orthogonal Frequency Division Multiplexing (OFDM) system.

Images